Facebook has embraced artificial intelligence (AI) and doubled staff numbers to tackle hate speech, fake accounts and suicide broadcasts on Facebook Live in order to make the platform safer.
“Facebook has used technology and tools with clearer community standards for its content to empower community-building for people and bring the world closer together,” said Simon Harari, public policy manager for content in Asia-Pacific at Facebook.
With its growing user base of 2.2 billion accounts, Facebook needs to build a “safer community, and a voice for people to express themselves, with fairness in the global community across different cultures," Mr Harari said.
Facebook recently published more details about its community standards, outlining what can and cannot be done on the platform.
Facebook's Content Policy team consists of experts in topics such as terrorism, hate speech and child safety in offices around the world.
The policy covers direct threats, bullying, attacks on public figures, violence, nudity, identity theft, dangerous organisations, sexual exploitation, hate speech, fraud and spam.
“We use technology and reports from communities to enforce our community standards,” Mr Harari said.
The report was also received by content reviewers around the globe, including Thais. The number of content reviewers has doubled from 10,000 early this year to 20,000 at present.
Moreover, Facebook has invested heavily in AI to proactively detect violent content.
In May this year, Facebook released the first Community Standard Enforcement Report for the last quarter of 2017 and the first quarter of this year to analyse six areas: violence, adult nudity and porn, terrorist propaganda, hate speech, spam, and fake accounts.
Hate speech covers race, religion, sex, ethnicity, serious disability, gender identity, national origin, caste and sexual orientation.
“The use of AI helps remove fake accounts and spam, but there are still some challenges in hate speech because in different cultural contexts, human content reviewers are necessary,” Mr Harari said.
AI has removed 583 million additional fake accounts, taking out 837 million spam posts, and discarded 21 million pieces of nudity and porn, 3.5 million instances of violent content, 1.9 million pieces of terrorist propaganda and 2.5 million instances of hate speech.
“Normally, we remove content or respond within 24 hours after receiving a notification," Mr Harari said. "In case of court warrants to remove content, Facebook has Community Standards and legal teams for each country.
“We worked hard to tackle these fake accounts in the past 18 months to reduce distribution of misinformation used in many government elections,” he said, adding that such misinformation can be exploited for financial incentives such as clickbait advertising or misleading voters.